Giorgio Veratti - Università di Modena

1. A MULTI-SCALE MODELLING
APPROACH FOR EVALUATION OF
URBAN AIR QUALITY IN
MODENA
Giorgio Veratti1, Sara Fabbi1, Alessandro Bigi1, Aurelia Lupascu2,
Gianni Tinarelli3, Sergio Teggi1, Giuseppe Brusasca3,
Tim M. Butler2 and Grazia Ghermandi1
1Department of Engineering “Enzo Ferrari”, University of Modena and Reggio Emilia, Modena, Italy
2Institute for Advanced Sustainability Studies, Potsdam, Germany
3ARIANET S.r.l., Milano, Italy

2. Goals of the project:
• Estimate the contribution of NOx traffic emissions in the city of Modena
through the application of multi-scale modelling approach
• Support local policymakers and inform citizenship about the quality of air
and the impact of urban emission sources, particularly traffic.
• Set-up a building-resolved air quality forecast system by a multi-scale model
approach

3. Multi-scale model flow chart

4. Case study
WRF-Chem Parallel Micro SWIFT SPRAY
Three one-way nested domains:
d01 -> resolution 15 km
d02 -> resolution 3 km
d03 -> resolution 1 km
Modena urban domain -> 6 km x 6 km
Resolution -> 4 m
Domain subdivided in 16 Tiles

5. WRF-Chem set-up
Main parametrizations Input parameters
Land-surface
model
Noah LSM Land-use Corine 2012
Boundary Layer
scheme
YSU
Meteorological
IC/BC
ECMWF Era-5
Gas-phase
mechanism
MOZART
Chemical
IC/BC
MOZART-4
Aerosol model MOSAIC 4 bins Vertical grid
35 levels
(30m ÷ 50 hPa)
Biogenic emissions: computed on-line with MEGAN v2
Anthropogenic emissions: TNO-MAC III emission inventory

6. (2) Downscaled inventory
(1) Original TNO MAC III emission
inventory
Modena urban traffic
emissions set to zero
(3)
longitude
longitude longitude
latitude
latitude
latitude
How can the double counting of NOx traffic emissions be avoided?

7. PMSS emissions estimation
Only traffic emissions were simulated with the PMSS modelling suite
Bottom-up approach
Traffic flows data on the main urban
road were estimated by a traffic
model: PTV VISUM
Output data at rush hour
(from 7:30 a.m. to 8:30 a.m.):
- Average travelling speed
- Number of duty vehicles
- Number of light vehicles
EMEP/EEA Cold & Hot Emissions Factors (Ntziachristos & Samaras, 2016)

8. PMSS emissions estimation
Hourly traffic modulation
Direct measurements of traffic flow have been continuously collected by four Doppler radar traffic
counters in the proximity of one the intersection with the urban ring road
(28 October – 8 November 2016)

9. Parallel Micro SWIFT SPRAY set-up
Micro-SWIFT Micro-SPRAY
Horizontal resolution 4 m Horizontal resolution 4 m
Horizontal grid 1504 x 1504 points Horizontal grid 1504 x 1504 points
Vertical grid from 3 up to 200m
20 vertical levels
Vertical grid from 3 up to 200m
10 vertical levels
Interpolation method Cressman 2D Emission time step 5 s
Fast momentum solver activated
Averaging period for
concentrations
3600 s
Simulation period
28 October – 8 November 2016
The same period whereby the traffic measurement campaign was carried out

10. MODELS EVALUATION RESULTS
WRF-Chem (3 km and 1 km resolution) combined with PMSS:
Variation of NOx concentration by hour of the day

11. MODELS EVALUATION RESULTS
WRF-Chem (3 km and 1 km resolution) combined with PMSS:
Statistics of hourly NOx concentrations
Configuration Station NMB FB NMSE FAC2 R
d02 + PMSS
Urban
traffic
-0.04 0.04 0.48 0.72 0.47
Urban
background
-0.25 0.29 0.75 0.62 0.44
d03 + PMSS
Urban
traffic
-0.15 0.16 0.54 0.68 0.48
Urban
background
-0.41 0.52 1.15 0.59 0.43
Reference acceptance criteria defined by Hanna and Chang* in urban environment
| FB | NMSE FAC2
< 0.67 < 6 > 0.30
*Hanna, S.R., Chang, J.C., 2012. Acceptance criteria for urban dispersion model evaluation. Meteorological Atmospheric Physics 116, 133–146

12. The full length paper can be temporary downloaded at the following link:
https://authors.elsevier.com/a/1aSOQ4pTZHdeVn

13. Hourly forecast of NO and NO2 (as well as their sum NOx)
up to one day ahead for the city of Modena
Simulation period
1 February – 28 February 2019
TESTING THE HYBRID MODELLING SYSTEM
(WRF-CHEM + PMSS) IN FORECAST MODE
WRF-Chem PMSS
2 nested domains 15km and 3km res. Traffic flow
Modelled data
(PTV VISUM )
Meteorological
IC/BC
GFS
Traffic modulation
Historical measured
data (230 sensors)
and cluster analysisChemical
IC/BC
WACCM

14. MODELS EVALUATION RESULTS
WRF-Chem combined with PMSS:
Variation of NO, NO2 and NOx concentration by hour of the day

15. MODELS EVALUATION RESULTS
WRF-Chem combined with PMSS:
Statistics of hourly concentrations
Station Pollutant NMB FB NMSE FAC2 R
Suggested benchmark values < 0.67 < 6 > 0.30
Urban
traffic
NO -0.01 0.02 1.50 0.42 0.42
NO2 -0.13 0.13 0.22 0.78 0.69
NOx -0.05 0.06 0.72 0.61 0.51
Urban
background
NO -0.44 0.57 3.30 0.23 0.15
NO2 -0.11 0.11 0.27 0.78 0.61
NOx -0.29 0.34 1.07 0.46 0.40

16. CONCLUSIONS
- The combination of WRF-Chem and PMSS modelling suite showed satisfactory
results in urban environment, with performances that fulfil the acceptance criteria for
urban environment regarding the FAC2, FB and NMSE metrics.
These result allow us to consider this approach as a good tool to support urban
mobility planning and policymakers
- Good results are also expressed by the hybrid modelling system applied in forecast
mode. Only NO concentrations at urban background site seem to be poorly
reproduced. It is worth noting that with this approach we are missing the secondary
contribution of the traffic in the urban environment.
- A comparison between the hybrid approach and the conventional Chemical
Transport modelling (WRF-Chem+PMSS vs. WRF-Chem) will be further investigated
in the next months.